J. Phys. Chem. 1995, 99, 4698-4709
4698
Characterization and Catalytic Studies on Defect Sites Formed upon the Thermal Decomposition of Sodium Ionic Clusters in NaX Zeolite Mark W. Simon? John C. Edwards: and Steven L. Suib**?s U-60, Department of Chemistry, University of Connecticut, Storrs, Connecticut 06269-3060; Texaco, Inc., Texaco Research and Development, P.O. Box 509, Beacon, New York 12509; and Department of Chemical Engineering and Institute of Materials Science, University of Connecticut, Storrs, Connecticut 06269-3060 Received: October 20, 1994@
Sodium ionic clusters have been prepared by chemical vapor deposition of sodium metal at 225 “C and 1 x Torr in NaX zeolites. A maximum loading of approximately 1.2 wt % sodium was vaporized onto dehydrated NaX zeolite. Ionic clusters prepared in this study have been determined to have the formula Na65+. In this complex the sodium ions are arranged in octahedral geometries in the sodalite cages of NaX zeolite. Identification of these species was accomplished by electron paramagnetic resonance, magic angle spinning nuclear magnetic resonance. Fourier transform infrared spectroscopy, luminescence spectroscopy, and X-ray diffraction experiments. These ionic clusters have a localized electron trapped in the center of the octahedra, giving rise to a color center. Infrared analysis of pyridine adsorption onto NaX and NaX exposed to sodium vapor (Na(v)/NaX) revealed that initial Bransted acidities of untreated NaX zeolites are low and that a complete loss of Bronsted acidity is observed upon treatment with Na vapor. The catalytic properties of materials exposed to sodium vapor have been studied in isomerization reactions. Defect sites or holes on framework oxygen sites are proposed as the source of catalytic activity at elevated temperatures (’340 “C). A catalytic mechanism for isomerization of cyclopropane to propylene over positively charged defect sites associated with Na65f clusters in Na(v)/NaX zeolites is discussed. Activation energies, based on Arrhenius plots for first-order reactions, and rates of propylene formation are determined for cyclopropane isomerization reactions over defect sites in Na(v)/NaX catalysts. Two reaction pathways are proposed for the treatment of sodium vapor on these materials. The dominating reaction produces N k 5 + clusters in the sodalite cages of NaX zeolites. In a second reaction, Na vapor reduces Bransted acid sites on terminal hydroxyl groups. The latter reaction leads to the formation of hydridic species on framework silicon atoms at high temperatures and the generation of electron holes on framework oxygen. Activation of defect sites has been observed only at high temperatures. Electron paramagnetic resonance studies show that naphthalene, sublimed on defect sites, results in the formation of naphthalene radicals.
I. Introduction
charge density confined to a small area of the zeolite. The small zeolite cages and accompanying negatively charged framework The use of sodium vapor to generate catalytically active make zeolites good host materials for highly charged ionic species in zeolites is highly unusual. Sodium and other metals clusters. Ionic clusters are stabilized inside zeolite cages due like vanadium, copper, and nickel, which are found in heavy to steric and electrostatic effects from the zeolite matrix. The hydrocarbon feed stocks, usually act as poisons in fluid cracking “protective” framework of the zeolite enables spectroscopicand catalysts.’ Several studies have shown that increased concentracatalytic studies to be carried out on these highly unusual and tions of sodium in zeolites are directly related to a decrease in reactive species. the catalytic activity of these material^.^,^ There are few detailed studies on the catalytic activity of ionic The use of zeolites for stabilizing highly reactive metal cluster^^.^ and ionic clusters, such as Nap-’)+ and K,(n-1)+,6-21clusters in zeolites.’ 1.16.17 Sodium ionic clusters, formed by the decomposition of sodium azide, have been studied in doublehas been of great recent interest. Exposure of zeolites with highbond isomerization of cis-2-butene and in the hydrogenation alumina contents to sodium metal vapor produces color centers of acetylene and benzene.17 Ionic sodium clusters have been which have been associated with ionic sodium clusters located prepared by the decomposition of NaN3I7-l9 and by photoirin zeolite cages6-I0 The properties of ionic clusters and studies r a d i a t i ~ n . ~y-irradiation ~.~~ of zeolites has been reported to on their formation have been of interest for synthetic and create solid-state defect sites resulting in materials with altered potential catalytic applications.6-’6 Absorption studies have also catalytic a ~ t i v i t y . * ~ . ~ ~ been investigated on NQ~$clusters in sodium sodalite materiMetallic clusters formed in zeolites have been of interest for als? their Ionic clusters have been reported to form in o ~ t a h e d r a l ~ ~ ~ ~ ~ . * ~catalytic properties in hydrogenation reactions and in reforming Fe-Mn mixed-metal clusters in or tetrahedral’3s16-23 geometries. The geometry of these highly zeolites have been studied as CO hydrogenation catalysts! Other charged clusters is unusual because of the existence of high small-particle bimetallic clusters of group VI11 (Fe, Co, Ni) metals have also been prepared in zeolite^.^ Bimetallic cluster * To whom correspondence should be sent. ’ Department of Chemistry, University of Connecticut. catalysts have been used in commercial reforming of petroleum Texaco, Inc. fractions and in improving selectivities for a number of 4 Department of Chemical Engineering and Institute of Materials Science, hydrocarbon reaction^.^^.^^ Small particle transition metal University of Connecticut. c? Abstract published in Advance ACS Absrructs, March 1, 1995. clusters formed in zeolites are highly stable in comparison to ~~
~
0022-3654/95/2099-4698$09.00/0 0 1995 American Chemical Society
Decomposition of Sodium Ionic Clusters in NaX Zeolite
1
Gku Frit
Figure 1. Chemical vapor deposition (CVD) reactor used in the application of sodium.
ionic clusters. Studies of the catalytic properties of ionic clusters, however, have been limited due to complicated synthetic procedures and their sensitivity to air and moisture.I6 This work focuses on the characterization of NaX zeolites chemically treated with sodium vapor. Resulting materials are used in cyclopropane isomerization reactions. Catalytically active sites have been studied by magic angle spinning nuclear magnetic resonance (MAS NMR), electron paramagnetic resonance (EPR), and Fourier transform infrared spectroscopy (FTIR) methods. EPR experiments of sublimed naphthalene on these materials were carried out to illustrate the presence of catalytically active defect sites in these materials. Stability of such materials is also studied by spectroscopic techniques. Mechanisms for the formation and thermal decomposition of ionic clusters and defects are mentioned. 11. Experimental Section
A. Synthetic Approaches. 1. Preparation of Sodium Ionic Clusters. About 0.25 g of 60 mesh NaX zeolite (Linde, from Alfa Ventron, Danvers, MA) was thoroughly dehydrated in a quartz tube at 380 "C for 15 h under vacuum (1 x Torr). Complete dehydration of these materials is imperative, because even trace amounts of water can effect product formation in the zeolite cages. The dehydrated sample was transferred into a drybox under an ultrahigh-purity N2 atmosphere and loaded into a quartz reactor, hereafter called a chemical vapor deposition (CVD) reactor, shown in Figure 1. The CVD reactor, having a diameter of 7 / 8 in., contains a drop tube outside of the furnace to hold the solid metal, an internal tube leading from the drop tube to the base of the reactor, a semipermeable glass frit to support the zeolite, and a 24/40 male adapter for connection to a vacuum line. Based on the formula for the unit cell of dehydrated Nf& zeolite, Na86A186Si1060384, about 1.2 wt % sodium was required to maximize the number of ionic clusters that could be formed in the zeolite. The amount of sodium used in these studies theoretically places one ionic cluster in each of the eight sodalite cages of each unit cell of the zeolite, assuming that one vaporized Nao atom leads to the formation of one ionic cluster. Taking this under consideration, an excess of sodium (about 25 mg) was added to the quartz tube in order
J. Phys. Chem., Vol. 99, No. 13, 1995 4699 to account for some loss of material in transfer and to compensate for any oxide species on the surface of the metal. The sodium metal was transferred into the drybox and added to the drop tube of the CVD reactor. The CVD reactor was sealed off and transferred back to the vacuum line, and the catalyst was heated to 380 "C for 1 h to ensure complete dehydration. The temperature was reduced to room temperature, and the sodium was introduced into the bottom of the reactor. The temperature was slowly raised to 225 "C, and the metal was vaporized onto the zeolite. The temperature was held at 225 "C until nearly all the metal was vaporized and until the color of the zeolite turned from white to dark blue (approximately 10 min, depending on heating rate). The reactor was cooled and transferred back to the drybox where it was sealed and stored for characterization and catalytic studies. 2. Sublimation of Naphthalene. Following CVD of sodium vapor, a small quantity of the material was treated with naphthalene. Naphthalene was sublimed onto the zeolite using a quartz tube on a vacuum line. The tube consists of a side arm which holds the naphthalene and a straight portion which holds the sodium-modified zeolite. Materials were loaded into the sublimation tube inside a drybox. The walls of the sublimation tube were heated with heat tape in order to prevent the naphthalene from condensing on the walls of the tube. A tube furnace was used to heat the naphthalene, and a liquid nitrogen dewar was used to condense the naphthalene onto the zeolite. About 10 mg of naphthalene was used to react with 25 mg of sodium-treated zeolite. After the sublimation tube was thoroughly evacuated, the vacuum valve was closed and the naphthalene heated to about 120 "C. Liquid N2 was applied only to the portion of the tube containing the zeolite. The sublimation was continued until all the naphthalene had vaporized. Naphthalene was sublimed on three variations of Na-modified materials: sodium-treated NaX, sodium-treated NaX which was heated to 380 "C under vacuum, and sodium-treated NaX which was heated to 380 "C in air. All materials contained the same amounts of sodium and naphthalene. The samples were brought into a drybox and sealed in quartz tubes for EPR analysis. B. Characterization. 1. Electron Paramagnetic Resonance. Electron paramagnetic resonance (EPR) measurements were carried out on a Varian E-3 ESR spectrometer at room temperature and 77 K near a frequency of 9.0 and 9.1 GHz with 100 kHz magnetic field modulation and a modulation amplitude of 4.0 G. 2 . Magnetic Angle Spinning Nuclear Magnetic Resonance (MAS NMR). MAS NMR experiments for 'H, 23Na,27Al,and 29Siwere done on a Varian Unity-300 superconducting NMR spectrometer system with full solids capacity. All spectra were acquired using a 7 mm, high-speed CP/MAS probe obtained from Doty Scientific, Inc. 'H MAS spectra were acquired at 299.95 MHz using a single-pulse experiment at 7 kHz MAS speed, using a 10" pulse tip angle and a 0.5 s recycle delay. Background 'H static spectra arising from probe materials were subtracted from all 'H spectra. The 29Sispectra were acquired at 59.59 MHz with a 4 kHz MAS speed, using a 30" pulse tip angle and a 5 s recycle delay. The 'H and 29Si spectra were referenced to tetramethylsilane (0 ppm). The 27Alspectra were acquired at 78.16 MHz, with a 7 kHz MAS speed, a 10" pulse tip angle, and a 0.2 s recycle delay. 27Alspectra were referenced to crystalline KAl(Si04)2*18H20(0 ppm). 23Na spectra were acquired at 79.35 MHz with a 5-7 kHz MAS speed, a 10" pulse tip angle, and a 0.5-1 s recycle delay. 23Na spectra were referenced to solid NaCl (0 ppm). All 27Al, 29Si, and 23Na spectra were acquired with a single-pulse experiment with high-
4700 J. Phys. Chem., Vol. 99, No. 13, 1995
power decoupling. Zirconia MAS rotors with tight-fitting end caps were used with the Doty probe. MAS NMR analyses were carried out on standard, untreated NaX zeolite, as well as samples of NaX treated with Na(v). The samples treated with Na(v) are the same as those studied by EPR and in catalytic reactions. Na(v)-treated samples were sealed in quartz tubes following the synthesis and loaded into the MAS NMR rotors under inert conditions in a glovebox to prevent exposure to air. The rotor end caps were tight fitting enough that no hydration of the sample occurred during the NMR experiment. 3. Fourier Transform Infrared Spectroscopy. FTIR spectra were collected on a Nicolet spectrometer with 2 cm-' resolution by using a deuterated triglycerine sulfate (DTGS) KBr detector. NaX zeolites treated with Na vapor were studied via infrared before and after exposure to air and before and after thermal treatment at 380 "C by using 15 mg (neat) pellets. Framework vibrations and acidity of NaX were also studied by infrared analysis. Brmsted and Lewis acidities of both Na-treated and untreated NaX zeolite were determined by quantitatively measuring chemisorbed pyridine in a manner previously described.29 4 . Luminescence. Zeolites exposed to sodium vpaor were sealed in an evacuated 2 mm diameter quartz tube for luminescence analysis. Luminescence excitation and emission spectra were recorded with a Spex Model 1680B 0.22 m doublemonochromator luminescence spectrophotometer. An Osram XBO 450 W xenon arc lamp was used for excitation. 5 . X-ray Powder Diffi-action. X-ray powder diffraction (XRD) experiments were done on a Scintag Model PDS 2000 diffractometer. Samples were mounted on glass slides by sprinkling powder on the slides to avoid preferential ordering. A beam voltage of 45 kV and a current of 40 mA were used. Samples were scanned from 5" to 60" 26. C . Catalysis. In a drybox, 150 mg of NaX treated with Na vapor was loaded into a stainless steel reactor having a length of 12 in and an i.d. of 5 mm. The catalyst was supported on a 200 mesh stainless steel screen and glass wool. The sample was treated in He (20 cm3/min) at 380 "C. A diagram of the reactor system showing the gas feed system, reactor, gas sampling valve, gas chromatograph, flow meters, and bubble meters can be found elsewhere.29 A thermal conductivity detector (TCD) was used on a 5580A Hewlett-Packard gas chromatograph with Poropak Q and T columns in series. The gas sampling system and calibration methods can be found elsewhere.29 The partial pressure of reactant was held constant by using a 1:10 mixture (9.09 mol %) of cyclopropane ( c - C ~ H in ~ )He in all experiments. Temperature was varied from 220 to 380 "C in order to determine apparent activation energies of reaction by using the Arrhenium equation as shown in eq 1, In k = ln(A) - (Ea/R)(l/n
(1)
where A is the preexponential factor, E, is the apparent activation energy, R is the ideal gas constant (82.057 cm3 atd(mo1 K)), and Tis the reaction temperature in kelvin. Rates of propylene formation were derived based on the following:
where P is the total pressure (1 atm), Qo is flow rate in m L / s (ambient), TOis the temperature (298 K), ~ c is ~the H partial ~
n/
Simon et al. g= 2.001
-
L l l l l l l l l l l l l l l l l l l 19 Lines
Figure 2. EPR spectrum of Na(v)/NaX zeolite at 77 K; scanned at a magnetic field, H,of 3250 f. 250 G, with a v of 9.13 GHz. g= 2.1 5
I\
1OOG
c--.,
g- 2.1 1
t
g= 2.0026
Figure 3. EPR spectrum at 77 K; scanned at a magnetic field, H,of 3300 i 500 G and a v of 9.13 GHz for Na(v)/NaX exposed to air.
pressure of propylene leaving the reactor, and W is weight of the catalyst(g). Rates of C3H6 formation are expressed as mol/ (s g). A 20% mixture of 1-butene (1-C4Hs) in Ar was prepared and studied in isomerization reactions over Na(v)/NaX materials. This analysis was carried out in a manner similar to c - C ~ H ~ isomerization reactions. Catalytic studies using 1-butene were carried out from room temperature to 100 "C. 111. Results
A. Characterization. I . EPR. The EPR spectrum of Na(v)/NaX, as prepared, is shown in Figure 2. This spectrum was taken at 77 K with a scan range of 3250 f 250 G and with a receiver gain of 2.5 x lo2. The sharp band at H = 3247.5 G has a corresponding g value of 2.001. A total of 19 smaller lines can be recognized between 3000 and 3500 G. The outer six lines of this spectrum are very weak in intensity and can be observed only upon expanding the spectrum off-scale. In addition, three of these lines are convoluted underneath the intense peak centered at 3247.5 G. This was confirmed by measuring a hyperfine splitting constant (A value) of 25.9 G. EPR spectra collected on Na(v)/NaX samples exposed to air did not show hyperfine splitting, but instead showed three peaks located at g = 2.0026, 2.0083, and 2.0780. This spectrum is shown in Figure 3 at a scan range of 3300 f 500 G. Upon heating this material to 380 "C, the EPR spectrum (Figure 4) shows a single intense peak at a g value of 2.0026. Two smaller, broad bands are observed at lower magnetic field. The g values of these bands are between 2.11 and 2.15. The hyperfine splitting structure of this material disappears upon heating at elevated temperatures.
Decomposition of Sodium Ionic Clusters in NaX Zeolite
J. Phys. Chem., Vol. 99, No. 13, 1995 4701
9- 2.0780
I\
B
600
t
Figure 4. EPR spectrum at 77 K; scanned at a magnetic field, H,of 3300 f 500 G and a v of 9.13 GHz for Na(v)/NaX heated to 380 "C.
4
2
0
-200 -400 -600 -800 -1000
Figure 6. 23Na MAS NMR spectra of (A) dehydrated NaX and (B) Na(v)/NaX zeolite.
200 6
200
Chemical Shift (ppm)
g= 2.0026
18 16 14 12 10 8
400
150
0 -2 -4 -6 -8 -10
Chemical Shift (ppm)
Figure 5. ]HMAS NMR spectra of (A) dehydrated NaX and (B) Na(v)/NaX zeolite.
Samples prepared with excess sodium showed a single sharp peak having a g value of 2.0013. The EPR spectrum of this material (not shown) reveals little evidence of hyperfine splitting. 2 . 'H,23Na, 27Al, and 29Si MAS NMR. 'H MAS NMR spectra for NaX and Na(v)/NaX are shown in Figure 5. In the untreated NaX zeolite, shown in Figure 5a, typical surface hydroxyl resonances (Si-OH) of NaX are observed at 1.8 ppm, AI-OH resonances are observed at 2-3 ppm, and bridging hydroxyl groups (Al-O(H)-Si) are apparent between 3 and 6 ppm. A signal of weaker intensity is observed at 7.8 ppm. The spectrum of Figure 5b, corresponding to Na(v)/NaX zeolite, shows a dramatic decrease in the 'H signal intensity of the AlO(H)-Si, AI-OH, and Si-OH resonances between 0 and 6 ppm. Surface hydroxyl groups at 1.8 ppm still present in Na(v)/NaX materials, however, are significantly reduced in concentration. A single, sharp signal is observed at approximately 0 ppm. A significant resonance is observed at -3 ppm. The resonance at 7.8 ppm in both materials is about the same order of intensity. The data of Figure 5 were plotted to full scale for each analysis; however, the difference in the two signals is that the intensity of the resonance of 7.8 ppm is much greater relative to the total hydroxyl contribution in sodium vapor-treated materials. The overall intensity of all hydroxyl groups in Na(v)/NaX materials has significantly decreased. This indicates that the species at 7.8 ppm may not be accessible to Na vapor, while hydroxyl groups are effected by sodium vapor. These
100
50
0
-50
-100
-150
Chemical Shift (ppm)
Figure 7. 27Al MAS NMR spectra of (A) dehydrated NaX and (B) Na(v)/NaX zeolite.
data strongly suggest that sodium vapor is highly reactive with hydroxyl groups of the zeolite. 23NaMAS NMR spectra for NaX and Na(v)/NaX are shown in Figure 6. The 23NaMAS NMR spectrum for NaX zeolite, shown in Figure 6a, reveals a broad asymmetric resonance due to several types of Na which are located between 0 and -150 ppm. This signal is surrounded by intensity due to spinning sidebands. Variable spinning speed experiments reveal the presence of at least four types of sodium contributing to the line shape. Broad resonances were also observed in samples heated at 380 "C (spectrum not shown). The 23Na NMR spectrum of Na(v)/NaX (Figure 6b) shows a much sharper, more symmetric signal with the majority of intensity at -12 ppm. 27AlMAS NMR spectra of NaX and Na(v)/NaX are shown in Figure 7. Figure 7a shows the 27AlNMR spectrum for NaX zeolites. This spectrum has a single, relatively broad peak located at 65 ppm. This peak is indicative of tetrahedral A1 located in the framework of the zeolite. Figure 7b shows a similar signal which also corresponds to tetrahedral aluminum. The similarity in the line shapes of these two samples will be discussed. The 29Si MAS NMR spectrum for NaX zeolite (Figure 8a) depicts four signals which are consistent with various Si sites typical of NaX zeolites. The Si:Al ratio of this material is about 1.25. The 29SiN M R for Na(v)-treated NaX is shown in Figure 8b. This spectrum reveals similar framework assignments for silicon between -80 and -110 ppm. The Si:Al ratio is about
Simon et al.
4702 J. Phys. Chem., Vol. 99, No. 13, 1995
-60
-70
-a0
-90
-100 -110
-120
-130 -140
4000
3000
3500
2000
2500
Chemical Shift (ppm)
Wavenumber, cm"
Figure 8. 29SiMAS NMR spectra of (A) dehydrated NaX and (B)
Figure 10. FTIR spectrum of Na(v)/NaX heated to 380 "C plotted from 4000 to 1800 cm-I.
Na(v)/NaX zeolite.
L
0 4000
3500
3000
2500
2000
1500
1000
Wavenumber, cm" Figure 9. Fourier transform infrared spectrum of Na(v)/NaX treated with pyridine for 1 h at room temperature.
1.4. This suggests that slight dealumination has occurred in materials treated with sodium vapor. 3. FTZR Data. The FTIR spectrum of Figure 9 corresponds to the analysis of adsorbed pyridine on materials containing sodium ionic clusters. The presence of a vibrational band at 3695 cm-' is observed in all sodium-modified zeolites that were not exposed to air or heat. This band is not observed in unmodified NaX zeolites. A broad band centered at 2000 cm-' is also observed. The spectrum of Figure 9 also shows a broad band between 3550 and 3000 cm-' which is indicative of the presence of water in this material. Water contamination is likely to have occurred upon transfer of the sample pellet into the infrared cell. The bands of Figure 9 at 1690, 1650, 1590, 1490, 1460, and 1365 cm-I correspond to physisorbed pyridine or Si-0 or A1-0 vibrational bands, indicating some deformation of the zeolite framework. The band at 1460 cm-' corresponds to Lewis acidity. Adsorption of pyridine on acid sites in unmodified NaX zeolite results in a very weak adsorption band at 1541 cm-I, consistent with Bronsted acidity, and a stronger band at 1460 cm-l, consistent with Lewis acidity. The FTIR spectrum of Na(v)/NaX materials (Figure 9) does not show any transmittance band at 1540 cm-', suggesting that loss of Bransted acidity occurs when NaX is treated with sodium vapor. The band at 1460 cm-' was observed to increase upon CVD treatments with sodium. The infrared spectrum of Na(v)/NaX heated to 380 "C is shown in Figure 10 from 4000 to 1800 cm-I. This spectrum
r
450
I
500
I
550
I
600
640
7b
7:O
ab0
Wavelength, nm Figure 11. Luminescence excitation and emission spectra of Na(v)/ = 475 nm, (B) I,, = 575 nm. NaX: (A) ,Icx
reveals an extremely intense band at 2208 cm-'. This band is, in most cases, the strongest observed in the spectrum and, to our knowledge, has never been observed in zeolites. This material shows no evidence of hydroxyl groups, indicative of either Bransted acidity or water. 4. Luminescence. The luminescence excitation and emission spectra of Na(v)/NaX are shown in Figure 11. This sample shows an excitation maximum, having a relatively strong intensity of 6 x lo4 counts per second (cps), for a species at a wavelength of 475 nm. An emission band corresponding to this excitation is observed at a wavelength of 575 nm. Untreated NaX zeolite was also analyzed by luminescence in the range 300-800 nm, and no excitation or emission bands were observed. 5. XRD. X-ray diffraction pattems were obtained for both NaX and Na(v)/NaX zeolites. A slight decrease in peak intensities was observed in the Na(v)/NaX zeolite, which had been exposed to air during analysis. Additional peaks corresponding to oxide species or sodium metal were not observed in the diffraction pattems of sodium-treated materials. This suggests that crystalline Na or Na20 is not present in sodiummodified materials or that particle sizes are smaller than 50 .& Na(v)/NaX samples heated to 380 "C show a decrease in overall intensity, but the relative intensity of all peaks appears to be the same. B. catalysis. Results of c-C& isomerization experiments are shown in Figure 12. These plots show the percent conversion and rates of propylene formation (mol/(s g) x lo6)
J. Phys. Chem., Vol. 99, No. 13, 1995 4703
Decomposition of Sodium Ionic Clusters in NaX Zeolite ~12.5
2om:: 'o 0200
250
300
350
400
"5:
50
Temperature, "C
Figure 12. c-C3H6 conversion (%) and rate of propylene formation (mol/(s 8)) x IO6 as a function of temperature for (A) NaX and (B) Na(v)/NaX zeolite.
inx lo3, K" Figure 13. Arrhenius plot of ln(rate) of C3H6 formation vs 1/T x lo3 (K)-'for (A) NaX and (B) Na(v)/NaX.
as a function of reaction temperature ("C) for both NaX and Na(v)/NaX. At 220 "C, both NaX and Na(v)/NaX have very low catalytic activities with conversions less than 1%. At 320 "C, however, Na(v)/NaX begins to show significant activity. An almost linear increase in reaction rate with increasing temperature is evident up to 380 "C, where nearly 90% conversion is observed. NaX zeolite, on the other hand, shows negligible activity ( < 5 % conversion) at 380 "C. The low activity of NaX can be attributed to trace amounts of Bransted acidity initially present in the zeolite. In addition, the activity of Na(v)/NaX increases with increasing time on stream, while the activity of NaX zeolite has been known to decrease with increasing reaction times.29 Figure 13 shows an Arrhenius plot of (UT) x lo3 K-' vs ln(rate) where the slope of the line is equal to -E,/R. From this plot, apparent energies of activation (Ea)for cyclopropane isomerization reactions over NaX and Na(v)/NaX are obtained. NaX has an Ea of 13.5 kcal/mol while Na(v)/NaX has a higher E, of 16.4 kcal/mol. 1-Butene reactions studied over Na(v)/NaX materials showed no evidence of isomerization to 2-butene at room temperature. Even at 100 "C a cis/trans ratio of 2-butene greater than the thermodynamic equilibrium value of approximately 1:2 was not observed. C . Naphthalene Sublimation. Figure 14 shows EPR spectra of naphthalene sublimed onto Na(v)/NaX zeolite. The spectrum
31 75
3200
3225
Magnetic Field (Gauss)
Figure 14. EPR spectra at 100 K of sublimed naphthalene on (A) Na(v)/NaX zeolite as prepared and (B) Na(v)/NaX heated to 380 "C under vacuum. Magnetic field scan range is 3187.5 & 37.5 G, with a v of 8.95 GHz.
shown in Figure 14a corresponds to naphthalene sublimed onto Na(v)/NaX zeolite, as prepared. Figure 14b shows the EPR spectrum of naphthalene sublimed onto Na(v)/NaX zeolite which was heated under vacuum to 380 "C. Figure 14a reveals a single sharp peak centered at a g value of 2.01 12. This spectrum does not show any splitting pattem which might be due to a naphthalene cation radical. Splitting due to the sodium ionic clusters has also been significantly reduced. Upon heating these materials the EPR spectrum (Figure 14b) shows a shift to a lower g value of 2.003. The shift to lower g values upon heating is consistent with EPR measurements taken on freshly prepared samples not containing naphthalene. The spectrum of naphthalene sublimed onto heated Na(v)/NaX samples shows peak broadening of the central peak and significant intensity due to hyperfine splitting.
IV. Discussion A. Sodium Ionic Clusters. Upon exposure to Na vapor at 225 "C for about 10 min, the dehydrated NaX zeolite turned from white to light blue and finally to dark blue or purple in color. Blue or purple color centers for NaX zeolites exposed to Nao vapor have been previously r e p ~ r t e d . ' ~ - ' ~ Others ,*~ report pink color centers for NaY zeolites exposed to Na vapor or NaN3.6,73"3'6330Pink color centers have been associated with Na3+ionic clusters, while blue color centers are observed with N G clusters. ~ ~ Upon exposure to air, these samples immediately tumed pale yellow or off-white in color. This is consistent with previous literature report^.^^,^^^^^ Samples treated with sodium for longer periods of time, i.e., samples having high concentrations of sodium, were black or gray in color, consistent with previous report^.'^.'^,'^ These samples exhibit little or no hyperfine splitting structure found in ionic clusters. When samples with high sodium concentrations are exposed to air, the black color faded over a period of several hours and remained gray in color. The gray color and slow diffusion rates of oxygen suggest that sodium metal clusters or particles are formed and block zeolite pores when excess sodium is used. The use of excess metal in the CVD of sodium was avoided for catalytic and spectroscopic analyses. Atomic absorption analysis shows a total sodium weight percent of 11.4%, which is equivalent to an excess of 1.2 wt % sodium added to the zeolite. From elemental analysis data, the net unit cell composition for sodium-treated NaX zeolite was determined to be (Nao)io(Na+)ssAlssSiio60~~4.
4704 J. Phys. Chem., Vol. 99, No. 13, 1995
Simon et al.
TABLE 1: Literature Data on g Values and Hyperfine Splitting Constants for Various Sodium Ionic Clusters in NaX and NaY Zeolites sodium ionic cluster zeolite Na3+ Na3+
Na3+ Na3+ Nas4+ Na5+ Nass+ Nass+ NkS+
NaX NaX NaY NaY NaX
NaX NaX NaX NaX
color and no. of lines purple/l3 purple/l3 red13 pinWl3 blue/l6 dark blue/l9 bluell9 purplell9 purple/l9
g value (lit.) A value (G) 2.0002 1.9990 2.0012 2.0012
29.0 28 32.5 32.5
a
2.0013 2.001 2.0022 2.001
a
25.9 21 25.0 25.9
ref 13 23 30 I 14b 12 20 14b this work
'Not reported.
B. Spectroscopic Techniques. 1. EPR. The intense asymmetric center line at g = 2.001 in the EPR spectrum for sodium ionic clusters has been assigned to small sodium particles in the cages of NaX,Io as pointed out by a reviewer. This is consistent with previous assignments for Na metallic particles and bulk metal which have been reported to have g values of 2.0013 and 2.0015, r e s p e c t i ~ e l y . ~The ~ . ~band ~ at g = 2.001 overlaps three other peaks as well as the central line splitting of the cluster itself. The g value of 2.001 and the hyperfine splitting constant of 25.9 G are both consistent with Na5+ ionic clusters reported for sodium-treated NaX zeolite^.'^^^^.^^ Table 1 shows a summary of g values and hyperfine splitting constants reported for various sodium ionic clusters in NaX and NaY zeolites. A total of 19 lines is expected to be observed for N a 5 + ionic clusters due to the contribution of six Na+ ions in an octahedral arrangement. This observation is based on the (2nI 1) rule with 1% = 3/2and n = 6 for each sodium ion in the cluster. In the EPR spectrum shown here, the central 13 lines are clearly observed; however, the outer six lines are extremely weak in intensity but are still present in these materials. In cases of low concentrations of Na65+ionic clusters, the intensity of the outer hyperfine splitting lines may not be within the detection limits of the instrument. In our EPR analyses, spectra for ionic clusters were expanded off-scale in order to detect the weaker, outer lines of the spectrum. An A value of 25.9 G results in 19 bands within a 492 G range. An octahedral environment is the most stable, least strained, and lowest energy arrangement of Na+ ions in the sodalite cages of the zeolite.22 Ionic clusters are believed to form from cationic sodium ions present in the zeolite, as shown in eq 3. The electron supplied to the cluster is believed to originate from the oxidation of sodium metal. These observations and the generalized equation expressed in (3) are consistent with previous reports.I0
+
6Na+
+ Nao - Na;' + Na'
(3)
The color center associated with the ionic clusters and a g value near that of a free electron (g = 2.0023) suggest that a highly energetic and localized unpaired electron is associated with the octahedra of sodium ions. The unpaired electron in Nq5+ clusters reported by Kevan et al. has been found to interact equally with all 27Alnuclei in the sodalite cage @-cage), as determined by Fourier transform electron spin echo modulation (FT-ESEM) experiment^.^^ The constrained environment of the sodalite cage and the high negative charge associated with the zeolite framework of NaX is likely to stabilize these highly energetic species.
Samples of NaX exposed to excess sodium resulted in black materials having an EPR spectrum with a g value of 2.0013. This is consistent with Nao particles greater than 1000 8, in size.31 Hyperfine splitting, indicative of ionic clusters, was not observed in these materials. This is consistent with studies by Anderson et al., who reported that the loss of hyperfine splitting in zeolites containing high concentrations of Nao could be attributed to interactions between two or more ionic clusters within 12 8, of each other.33 Sodium ions of clusters in adjacent sodalite cages are approximately 5 8, apart. This is based on geometric calculations on the clusters and the free pore diameter of the zeolite. Electron hopping between ionic clusters of adjacent sodalite cages has also been suggested.33 An additional reaction that occurs when NaX zeolites are exposed to sodium vapor is represented by eq 4. This mechanism accounts for the loss of Brosted acidity as a result of treatment with sodium vapor and implies that there is a 1:1 stoichiometry between Bronsted sites (H+zeolite)and ionic clusters. This mechanism shows that resultant electron holes (h+) are present in sodium-treatedmaterials. These holes result from the donation of an electron from framework oxygen of the zeolite to the N a 5 + cluster, which, in tum, stabilizes the highly charged complex in the sodalite cage of the zeolite. A g value of 2.0123 has previously been assigned to holes in N&Y and HY zeolite^.^^,^^ The loss of Bransted acidity may result in the generation of hydrogen radicals as shown in the redox reaction of eq 4; or proton loss may occur in the form of 5Na'
+ Hfzeolite
Na;'
+ hf + H'
(4)
molecular hydrogen. We have no evidence for the generation of H2 during the reduction of Bronsted acidity. Further, the amount of Bronsted acidity present in these materials is very low; therefore, the production of H2 or atomic hydrogen will be limited. Although the presence of Bronsted acidity is conceivably not necessary for ionic cluster formation, the mechanism of eq 4 justifies the loss of Bransted acidity during reaction with sodium vapor. It has been suggested by a reviewer that the generation of electron holes may very well be a result of direct reaction between surface hydroxyl groups and sodium metal and may not involve cluster formation at all. The production of atomic hydrogen and the reduction of Bronsted acidity are consistent with lTIR and pyridine chemisorption experiments which will be discussed later. The EPR spectrum of ionic clusters exposed to air shows an intense signal at g = 2.0026, a shoulder at g = 2.0083, and a weaker signal at g = 2.0780, which are consistent with oxygen radicals. Specifically, the data of Figure 3 suggest that the superoxide anion, 0 2 - , is the EPR-active species in these materials. This is consistent with previous report^.^^,^^ Irradiated zeolites exposed to air at room temperature also show three g tensors in the EPR spectrum taken at 77 K.34 These g values, gxx = 2.0026, g,, = 2.0085, and g,, = 2.0575, have been observed in N&Y zeolites.34 In NaX zeolites, g,, = 2.078 has been assigned to a 0 2 - superoxide species adsorbed on Na+.35 The absence of hyperfine structure in this sample suggests that there is no interaction between 0 2 - and electron holes associated with aluminum.34 The data of Figure 4 show a loss of fine structure and a slight shift in g values upon heating at 380 "C,which suggests that the decomposition of ionic clusters occurs at high temperatures. This is consistent with previous report^.^^^^ The EPR spectrum obtained for this material is very similar to those reported for the superoxide anion attached to a nearby Na' ~ a t i o n . ~The .~~ decay of the band at gz, = 2.078 into a broad band containing
Decomposition of Sodium Ionic Clusters in NaX Zeolite
J. Phys. Chem., Vol. 99, No. 13, 1995 4705
TABLE 2: Experimental and Literature Data of g Values for Sodium-Treated Materials sodium treatment color g value (exptl) g value (lit.) assignt ref low sodium loading Dude 2.001 2.001 ionic clusters 20,33 low sodium loading off-white 2.0026 35 2.0016 surface 0 2 2.0066 2.0083 2.0078 6 2.0078 defect site on framework 0 exposed to air white 2.0026 2.003 02--Na+ 36,31 low sodium loading heated at 380 O C in air 2.113 surface 0 2 35 2.11 -2.15 2.0013 Na particles 31 excess sodium loading grayblack 2.0013 2.0015 bulk Na metal 32 several new components between g, = 2.11 and 2.15 is in EPR studies.39 Solid state defect sites near g = 2.003 have been found to readily produce the 3,3-dimethyl-1-butene radical consistent with a previous study for samples irradiated at 77 K cation in ZSM-5 zeolites.39 Radical cations prepared at low that were warmed.36 Broadening of the g,, component is also temperatures over defect sites, however, are usually limited to likely to occur in these systems due to strong dipolar interactions organic molecules with an ionization energy less than 9.0 eV.40 when atomic hydrogen combines with 02-.36 The generation Naphthalene has an ionization energy of 8.14 eV.4' Intramoof atomic hydrogen in these systems and its consequences will lecular rearrangement reactions and reactions which generate be discussed later in more detail. The sharp band at g = 2.0026 carbenium ions are known to occur over positive holes!2 The may also correspond to a surface defect site, which has been generation of hexamethylbenzene radical cations over positively reported to have a value of 2.003.37 Differences in EPR spectra charged holes has also been studied by EPR.43 The reaction of between freshly prepared materials at 225 "C and materials aromatics and alkenes over positively charged defect sites and heated to 380 "C suggest that the types of defect site present V-centers is ~ e l l - k n o w n . ~ ~ . ~ ~ ~ ~ ~ following heating are quite different from holes generated in the formation of the sodium ionic clusters. Heating of these These EPR data suggest that Na(v)/NaX materials heated to materials is likely to change the environment surrounding defect 380 "C are reactive toward naphthalene while those materials sites and may cause migration of holes from one site to another. not thermally activated are unreactive toward naphthalene. These The formation of 0 2 - species can be attributed to the thermal observations suggest that positively charged defect sites exist breakdown of the N%5+clusters. The source of oxygen is likely in these materials in order to maintain charge neutrality; to be from the release of strongly adsorbed oxygen from the however, the type of defect site or its surrounding environment zeolite which is known to be released at high temperat~res.~, is different than that of holes produced during cluster formation. Thermal treatment results in greater interactions between trapped This is consistent with EPR measurements taken on freshly electrons and surrounding nuclei,33which might release trapped modified zeolites which show a remarkable difference in EPR electrons. The decomposition of ionic clusters at high tempersignals before and after heating. FTIR data also show a atures results in the release of the unpaired electron, which was significant difference in spectra taken before and after heating once associated with the cluster. The electron is then trapped Na(v)/NaX materials. There are far less adsorbed water species by gas phase oxygen molecules (trapped in the zeolite), which after heating Na(v)/NaX (Figure 10) than before heating (Figure serve as an electron sink. It might be expected that electrons 9). Electron holes may tend to increase the amounts of adsorbed would quickly recombine with positive centered holes; however, water prior to heating. it is well-known that cations adsorb 0 2 molecules, and these Heating of some Na(v)/NaX materials may lead to sintering species serve as excellent electron traps, giving rise to EPR of Nao particles into (Na)x sodium agglomerate^.'^,'^ EPR and signals characteristic of the superoxide ion.6,35 This process XRD results for our materials with low levels of sodium do suggests that active defect hole sites are produced after not show this trend. The presence of oxygen in these materials decomposition of Na65+ clusters. Table 2 shows a summary of is likely to result in oxidation of sodium metal upon decomposiexperimental g values for modified Na(v)/NaX materials along tion of the ionic clusters, particularly at high temperatures. with assignments and literature values. However, samples containing high levels of sodium (samples We have studied the reaction of naphthalene with ionic black in color) show evidence of particle migration and clusters and holes by means of EPR. Naphthalene was chosen agglomeration from EPR data, giving rise to an intense signal as a probe molecule for defect sites due to the distinct hyperfine for Na metal at g = 2.0013, consistent with agglomerates splitting structure of the naphthalene radical. The data of Figure sodium.28 14a show no evidence of peak broadening between 3175 and 2 . MAS NMR. The 'HMAS NMR spectra of Figure 5 show 3225 G that would result from the hyperfine splitting contribusignificant differences for NaX zeolite and NaX treated with tion of a naphthalene radical species. This suggests that radical sodium vapor. The peak at 7.8 ppm can be attributed to a cations are not formed in freshly prepared ionic cluster materials nonbonded hydronium ion as previously assigneda or to and that the ionic clusters themselves are not reactive toward strongly adsorbed water on Lewis sites in the zeolite, which naphthalene. The EPR spectrum of Figure 14b, however, is has been assigned a chemical shift of 6.5 The nature of quite similar to that of the naphthalene cation radical. This this Lewis site may effect its chemical shift values. The radical cation shows peak broadening as in Figure 14b, which intensity of the resonance at 7.8 ppm remains strong, indicating is about 34 G. The hyperfine splitting of this sample has an A that it is a site not effected by sodium vapor. The sharpening value of 1.4 G, also consistent with the naphthalene radical. of this resonance following treatment with sodium vapor The g value of the naphthalene cation is also near 2.00. Both suggests that either sodium vapor selectively reacts with peak broadening and fine structure observed in Figure 14b are hydroxyl sites but not with the site at 7.8 ppm or that a presumably the result of oxidation of naphthalene over positively completely new type of site is formed at 7.8 ppm as a result of charged defect sites on these materials as shown in eq 5. sodium vapor treatment. The peak at 0 ppm or the broader resonances at -3 ppm can be attributed to hydridic species in the zeolite. These resonances are in the hydride chemical shift range; however, there are no previous reports for assignments Positive holes on oxygen atoms have been previously of hydridic species in zeolites. A sharp decrease in Si-O(H)investigated using 3,3-dimethyl-l-butene as a probe molecule Y
A
I
4706 J. Phys. Chem., Vol. 99, No. 13, 1995 A1 signals and a significant decrease in Si-OH and A1-OH resonances upon treatment with Na vapor is strong evidence that the reactive species with Nao are bridging and terminal hydroxyl groups. Redox reactions of Bransted acid sites with Na vapor are ~ e l l - k n o w n . ~ ~ The four components in the 23NaNMR line shape of NaX (Figure 6a) correspond to Na+ located at various sites (I, 1', 111, and 111) throughout the NaX zeolite framework. Most Na+ ions reside in site I and I' in the sodalite cages, while smaller amounts of Na+ ions are located in the supercages. In NaX zeolite one observes signal intensity due to Na+ in site I (hexagonal prism site) which gives rise to intensity at -12 ppm. The broader resonances in the -20 to -50 ppm region are due to a convolution of resonances from Na+ in low symmetry sites 11, 111, and I'!6,47 In this particular sample the broader line shapes are due to asymmetric II, 111, and I' sites for Na+ that dominate the spectrum. In the case of the Na(v)/NaX sample, one does not observe the same Na+ distribution. The dominant feature of the spectrum (Figure 6b) is due to Na+ in hexagonal prism sites. This resonance is narrow owing to the highly symmetric octahedral environment of the Na+ ions. A dramatic loss of signal intensity after Na(v) treatment is observed in the -20 to -50 ppm range. This loss of signal intensity can be attributed to the presence of N G ~ +clusters with trapped electrons within 10 %, of the Na+ ions that are present in the asymmetric 1', 11, and I11 sites. Thus, it appears that the Nas5+ cluster is present in close proximity to only five of the six Na+ sites in the zeolites. Finally, metallic NaO is not observed in the 23Naspectra of the Na(v)-modified zeolites. This may be an indication that NaO is also closely associated with the N G ~ +clusters which quench the NMR signals due to electron-nuclear dipole interactions. 27AlNMR shows the presence of only framework tetrahedral aluminum, with no observation of extraframework aluminum. However, 29SiNMR reveals a slight dealumination of the NaX zeolite by the sodium vapor modification. The 29Siresults do not reveal the presence of unusual surface Si species; however, this may be due to a low concentration of these species, preventing observation of these species at the signal-to-noise levels of these experiments. 3. FTIR. Infrared analyses of chemisorbed pyridine indicate that some initial Bransted acidity is present in NaX zeolites. Infrared studies on Na(v)/NaX zeolites clearly show that Bransted acidity is eliminated upon treatment with Na vapor, as illustrated in eq 4. This is consistent with earlier work on ion-exchanged Eu3+/NaX, which showed that high initial Bransted acidities were eliminated when exposed to sodium vapor.29 EPR data for Eu3+/NaX exposed to Na vapor show a reduction of Eu3+ ions to E U ~ + ! ~This is significant because it suggests that sodium vapor destroys the (Eu4O)I0+complex, which has been reported to form in the sodalite cages of NaX upon d e h y d r a t i ~ n . ~ ~ ? ~ ~ Lewis acidity is still present in materials treated with Na vapor. An increase in pyridine uptake on Lewis sites in Na(v)/NaX samples is observed. This suggests that defect sites may also adsorb pyridine. It seems unlikely, however, that a single electron vacancy would permit a chemisorptive bond between pyridine and framework oxygen. The observation of increased pyridine uptake, therefore, is unexplainable by posing a single electron vacancy. If a two-electron vacancy is characteristic of the defect site, increased pyridine uptake would be expected. However, there is no way of knowing whether such species are present in these materials, for this type of site would be diamagnetic and EPR inactive. However, it is known that water reacts with these type of materials and produces a
Simon et al. diamagnetic species.I3 Therefore, at present, no firm conclusions can be made about the exact nature of these defects. EPR and FTIR data before and after heating suggest that the nature of the holes and surrounding environment are quite different following heating. The band at 3695 cm-' observed in Na(v)/NaX zeolites (Figure 9) is significant in materials having N G ~ +clusters. This band is indicative of water molecules hydrogen bonded to framework oxygen ions that were once associated with strong acidity and that the interactions between the free proton of hydrogen-bonded water and Na+ ions give rise to this cationdependent band at 3695 c ~ - ' . ~ O These data are also consistent with the interaction of Na+ with adsorbed water on the framework hole. This evidence suggests the presence of holes, and this band has been assigned to the presence of such defect sites. The presence of water in these materials is evident in the OH stretching region of the infrared. Infrared analyses definitively show the presence of silicon hydride (Si-H) on the surface of the zeolite. The peak at 2208 cm-' has been assigned to Si-H. A value of 2260 cm-I has been reported for the formation of silicon hydride on the surface of alumina by a silanization process.51 The assignment of this peak can easily be made because it is located in a very specific, characteristic region in the IR where only hydrides, cyanides, and carbynes are located. The latter two functional groups are not possible in these materials. Aluminum hydrides are known to occur at 1700 ~ m - ' .'H ~ MAS ~ N M R data are also consistent with the formation of a Si-H species. Hydridic species might be expected to form in the presence of H' species generated upon reaction of Bransted sites or water with metallic sodium, as shown in eq 4. In addition, a reduction in the zeolite framework by NaO, as evidenced by MAS NMR and X-ray diffraction, might facilitate the formation of Si-H species by creating available bonding sites on silicon. The formation of hydridic species is enhanced with increasing temperature. Upon an increase in temperature, the band at 2208 cm-' becomes sharper, more intense, and increasingly welldefined. This is evident by comparing the spectra of Figures 9 and 10. The source of atomic hydrogen has been suggested to come from the reaction of Na vapor with Bransted sites or from residual water in the zeolite,23although spectroscopic evidence for the formation of hydrides in zeolites has yet to be reported in ionic cluster formation reactions. In addition, it has been reported that H'does not react with V-center holes, generated during cluster d e c o m p ~ s i t i o n . ~ ~ These observations imply that the formation of silicon hydride during thermal decomposition of the clusters is a probable reaction. Sodium oxide might also form upon heating the Na cluster system. This may occur in the presence of oxygen at high ('100 "C) temperatures. Release of oxygen from such zeolites is known to occur at high temperature^.^^ The modification of the zeolite structure by formation of hydridic species on the surface is likely to result in surface defect sites. The EPR signal at g = 2.003 in heated materials may correspond to such a defect site. A reduction of the zeolite framework may result either from metallic sodium or through the breakdown of ionic clusters. Both materials are easily oxidized; therefore, it is difficult to determine which species is responsible for the hydride reaction or if both species are acting synergistically with each other in this process. Luminescence spectra of Figure 11 confirm the presence of an emitting species. We have assigned this species to a trapped electron in the N Q ~ +cluster. The transition observed in the luminescence might be due to interactions between two (or
Decomposition of Sodium Ionic Clusters in NaX Zeolite more) ionic clusters located in adjacent sodalite cages, which has been suggested to or between the electron and surrounding nuclei. This subject was previously discussed (see discussion on EPR). A blue center has been associated with a highly energetic species and, more specifically, in similar metal ion clusters as trapped electrons. The color center and luminescence are lost upon heating these materials at 380 "C. This suggests that the cluster decomposes at higher temperatures and that the trapped electron is released from the cluster. C . Catalysis. Catalytic and spectroscopic experiments suggest that the degradation of N a 5 + ionic clusters in NaX zeolites greatly enhances catalytic cracking properties at high temperature. It is unlikely, however, that the clusters themselves are catalytically active. The negatively charged (formal charge) N a 5 + clusters are unlikely to cause ring opening since literature studies suggest that ring opening occurs on positively charged site^.^^,^^ Furthermore, N a 5 + clusters are reported to exist only in the sodalite where cyclopropane cannot diffuse, based on size constraints of the molecule.49 Finally, the clusters decompose before the onset of isomerization as shown by spectroscopic and catalytic studies. Upon exposure to air or moisture for an extended period of time (> 10 min), Na(v)/NaX loses its catalytic activity. This is consistent with data of Martens et al., who report that catalytic activity of such systems is extremely sensitive to oxygen and water molecules.'6 Butene has been reported to isomerize to 2-butene with a high cisltrans ratio at 300 K over basic materials such as N a ~ 0 . 5 ~ Butene isomerization experiments studied here show that no reaction is observed up to 100 "C over Na(v)/NaX materials. This suggests that Na2O is not the catalytically active site in this reaction. Furthermore, we know of no mechanism for the base catalysis of cyclopropane. c - C ~ Hring ~ opening reactions occur through a carbenium ion intermediate and are known to occur typically over acid site^,^^,^^ not over base sites. In addition, previous experiments showed that activity in cyclopropane isomerization reactions is drastically diminished upon exposure of acid zeolites to sodium vapor.29 Sodium metallic particles have also been reported to be the active sites in the base catalysis of Recently, Martens et al.55 investigated zeolites containing sodium clusters as base catalysts for butene isomerization. From this study, it was suggested that framework oxygen anions in the neighborhood of intercrystalline neutral sodium clusters act as the active base site. Spectroscopic results form our studies suggest that 0 2 - anions do not react with framework defect sites. Furthermore, butene isomerization reactions over sodiumtreated materials prepared here show no catalytic activity at 300 K whatsoever. Our results suggest that base catalysis is not a contributing factor in zeolites modified by sodium vapor. Further we have shown that zeolites containing excess sodium have diminished catalytic activities in C-C& isomerization reactions. Although this may be a result of pore blockage, it also shows that sodium metal on the surface of the crystallite is inactive in c-C& isomerization reactions. Adsorption of cyclopropane on negatively charged centers such as hydrides is unlikely due to the fact that cyclopropane has high electron charge density in the center of the ring>6 Previous mechanisms suggest that adsorption and reaction of C-C& to C3H6 occurs over positively charged Bransted site~.*~%53 Iron impurities have been suspected to be catalytically active sites in zeolite^.^' The complete reduction of Fe3+ to Feo with sodium occurs at high temperature and with the use of excess
J. Phys. Chem., Vol. 99, No. 13, 1995 4107 sodium.57 EPR data show that FeO is not observed in our systems, suggesting that Na vapor treatment conditions used here are relatively mild. This is consistent with XRD analyses and 27AlMAS NMR studies which showed only slight degrees of dealumination of these materials. Furthermore, inductively coupled plasma analyses on NaX starting materials show less than 250 ppm Fe3+ impurity in these materials. These trace amounts of Fe3+ are unlikely to account for the high propylene conversions that were observed. Partial reduction of Fe3+ to Fe2+ can occur at lower temperatures as shown in eq 6. Fe3+ f Nao
-
+
Fe2+ Na+
(6)
Certain transition metal ions are thought to be active sites for the isomerization of C-C& at higher temperatures5* When excess sodium is applied, it is expected that reaction 6 would be driven to the right. If Fez+ species were present and catalytically active, then the activity would be expected to increase with increasing sodium concentrations. Catalytic data suggest, however, that excess sodium causes complete poisoning of catalytically active sites, resulting in a total loss of catalytic activity. The most plausible explanation for increased catalytic activity at high temperatures, therefore, is that defect sites created upon treatment with Na vapor or holes associated with N a 5 + clusters catalyze c - C ~ Hisomerization ~ reactions. FTIR and NMR analyses show that desorption of tightly adsorbed water molecules to holes occurs upon heating at high temperatures. The desorption process results in an EPR active band at g = 2.003. Water associated with framework holes, prior to dehydration, is known to produce a diamagnetic material.I3 The complete desorption of adsorbed water (as evidenced by FTIR) frees up catalytically active holes at high temperatures and provides an oxidative center available for hydrocarbon conversion reactions. This is consistent with both the oxidation of naphthalene and the isomerization of cyclopropane on thermally treated Na(v)/NaX materials. Holes on framework oxygen are stabilized between the aluminum and silicon atoms of the zeolite. Lewis sites associated with framework aluminum can further stabilize these defects. Substantial spectroscopic evidence from our studies suggests that the recombination of electrons (released during the thermal decomposition of ionic clusters) with framework holes does not occur at high temperature. The presence of 0 2 species suggests that it is more favorable for electrons to combine with molecular oxygen trapped in the zeolite than with the holes. This may be a result of the close proximity of free electrons to oxygen or due to a greater accessibility of mobile gas phase oxygen to electrons rather than to the holes on framework oxygen. The thermal generation of superoxide anions is energetically favorable. The spectroscopically characterized species generated upon heating of sodium ionic clusters support the existence of positively charged holes at high temperatures. The generation of defect sites has been previously observed in the formation of ionic clusters via y - i r r a d i a t i ~ n . ~ The ~.~~.~~ physicochemical changes observed in samples exposed to y-irradiation have been explained in terms of trapped electrons, holes, lattice vacancies, and interstitial entities.59 Changes in adsorption properties and catalytic activity which result from irradiation have been explained by two mechanismsm One theory involves a structurally unmodified system in which holes are formed as a result of irradiation which ionizes the surface of the material. The second involves a structurally modified system whereby y-irradiation generates defect sites
Simon et al.
4708 J. Phys. Chem., Vol. 99, No. 13, 1995
on the surface of the material. In zeolites modified by sodium vapor, it appears that both theories might apply. In our study, in the formation of positively charged holes, it is apparent that sodium ionic clusters are acting as reducing agents. Sodium can act as a reducing agent which can result in structural modification of the zeolite by creating a hydride species on the zeolite. We suggest that the formation of Si-H species is likely to result in the presence of defect sites possibly associated with the hydride. Catalytically active holes and defect sites have been previously r e p ~ r t e d . ~The ~ . ~concentration ~.~~ of induced defect sites on aluminosilicate frameworks has been directly related to catalytic activity in cumene cracking.62 A charge transfer reaction has been invoked which involves the generation of carbenium ions over positively charged holes.39 Similarly, the ring opening of conjugated radical cation systems has been attributed to positively charged holes.40 The catalytic isomerization of C-C& in our system is a result of positively charged holes that alter the adsorptive and catalytic properties of zeolites. Differences in activation energies between NaX and Na(v)/ NaX may be explained in terms of differences in mechanisms resulting from the presence of different types of active sites in the two systems. It is clear that activity in NaX is attributed to low initial Bronsted acidity present in these materials. Large E, values for Na(v)/NaX may be a result of stronger interactions of cyclopropane with holes. C-C& may chemisorb more strongly to the holes, thereby increasing the heats of sorption, reaction, and desorption, which result in increased activation energies. Higher reaction temperatures required to activate Na(v)/NaX catalysts are also consistent with higher activation energies. Deactivation of NaX occurs rapidly within the first 30 min time on stream, whereas the Na(v)/NaX catalysts do not deactivate at higher temperatures. This suggests that catalytic intermediates formed on the surface readily desorb on Na(v)/ NaX, whereas intermediates may react further and oligomerize or polymerize on the Brtinsted sites of NaX.
and resulting defects, are not acidic. Lewis site adsorption significantly increases for Na vapor-modified materials. FTIR data also suggest the presence of a strong adsorption site in these materials that has been assigned to holes in the material. Catalytic results for cyclopropane isomerization reactions show that conversion to propylene increases from 5% to 85% in materials containing N%5+clusters. Catalytic activity in the supercages of Na(v)/NaX has been associated with the formation of N%5f ionic clusters. Ionic clusters themselves have been shown not to be catalytic centers. Positively charged holes or defect sites, formed in the zeolite upon treatment with sodium, have been suggested to be responsible for C-C& ring opening. EPR experiments of sublimed naphthalene on sodium-treated NaX materials suggest that positively charged defect sites formed upon thermal treatment of ionic clusters are different than those produced upon ionic cluster formation. Catalytically active sites are produced at high temperatures. Adsorption of water to holes following cluster formation prevents catalytic activity. Thermal activation of holes changes the environment surrounding these defect sites and results in catalytic oxidative centers on Na(v)/NaX materials. Treatment of NaX zeolite with excess sodium resulted in a total loss of catalytic activity. EPR signals corresponding to sodium ionic clusters or holes were not observed in samples prepared with excess sodium. Trace amounts of iron impurities also do not appear to be responsible for catalytic activity. Activation energies for C-C3H6 reactions are approximately 3 kcal/mol greater for Na(v)/NaX than NaX. This has been explained in terms of stronger interactions that might exist between adsorption sites in Na vapor-modified zeolites and cyclopropane.
Acknowledgment The authors thank Roberto N. DeGuzman for collecting EPR data. We also thank Dr. Lennox E. Iton at Argonne National Laboratory for helpful discussions. We gratefully acknowledge the Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences and Texaco, Inc., for support of this research.
V. Conclusions Ionic sodium clusters having the formula N a 5 + have been produced in NaX zeolites by a CVD process from sodium metal. Nas5+clusters have been determined to be present in Na vaportreated NaX zeolite and characterized by EPR, MAS NMR, FTIR, and luminescence. EPR signals for N%5+clusters exhibit hyperfine structure and g values characteristic of holes on framework oxygen atoms of the zeolite are present in these materials. Exposure of N;b5+clusters to air results in three EPR bands which have been assigned to the superoxide ion adsorbed on Na+ cations. Heating of Na65f clusters results in decomposition of the clusters and an EPR signal corresponding to 0 2 - . Oxygen serves as an electron trap, preventing the free electron from recombining with existing holes upon decomposition of the clusters. Heating of ionic clusters results in structural modification of the zeolite surface. NMR and FTIR results suggest the formation of Si-H species on the surface of the zeolite, particularly at higher temperatures. This structural modification may lead to defect sites on the catalyst. Structural modification, in turn,may change the environment, properties and the location of framework defect sites. 23NaNMR reveals the presence of N%5+ clusters in close proximity to Na+ in sites 1’, 11, and 111, but not in site I. Metallic sodium also appears to be closely associated with N%5+ ionic clusters. Pyridine adsorption studies have shown that zeolites exposed to sodium vapor lose Brosted acidity and that ionic clusters,
References and Notes (1) Bhatia, S.; Beltramini, J.; Do, D. D. Catal. Rev.-Sci. Eng. 1989, 31, 455. ( 2 ) Fritz, P. 0.;Lunsford, J. H. J . Catal. 1989, 118, 85. (3) Beyerlein, R. A.; McVicker, G. B.; Yacullo, L. N.; Zeiemiak, J. J. Proc. ACS Meet. Div. Petrol. Chem., New York 1986, 31, 190. (4) Barrault, J.; Renard, C. Appl. Catal. 1985, 14, 133. ( 5 ) Sinfelt, J. H.; Carter, J. L.; Yates, D. J. C. J. Catal. 1972, 24,283. (6) Kasai, P. H. J . Chem. Phys. 1965, 43, 3322. (7) Rabo, J. A.; Angell, C. L.; Kasai, P. H.; Schomaker, V. Discuss. Faraday SOC.1966, 41, 328. (8) Ben Taarit, Y.; Naccache, C.; Che, M.; Tench, A. J. J. Chem. Phys. Lett. 1973, 24, 41. (9) Westphal, U.; Geismar, G. Z . Anorg. A&. Chem. 1984, 508, 165. (10) Hanison, M. R.; Edwards, P. P.; Klinowski, J.; Thomas, J. M. J . Solid State Chem. 1984, 54, 330. (11) Martens, L. R. M.; Grobet, P. J.; Vermeiren, W. J. M.; Jacobs, P. A. Stud. Surf Sci. Catal. 1987, 31, 531. (12) XU, B.; Kevan, L. J. Chem. SOC.,Faraday Trans. 1991, 87, 2843. (13) Yoon, K. B.; Kochi, J. K. J . Chem. SOC., Chem. Commun. 1991, 510. (14) (a) Anderson, P. A.; Singer, R. J.; Edwards, P. P. J. Chem. SOC., Chem. Commun. 1991, 914. (b) Anderson, P. A,; Edwards, P. P. J. Chem. SOC.,Chem. Commun. 1991, 915. (15) Edwards, P. P.; Hanison, M. R.; Klinowski, J.; Ramdas, S.; Thomas, J. M.: Johnson, D. C.; Page, C. J. J. Chem. SOC.,Chem. Commun. 1984, 982. (16) Martens, L. R. M.; Grobet, P. J.; Jacobs, P. A. Nature 1985, 315, 568. (17) Fejes, P.; Hannus, I.; Kiricsi, I.; Varga, K. Acta Phys. Chem. 1978, 24, 119.
Decomposition of Sodium Ionic Clusters in NaX Zeolite (18) Fejes, P.; Kiricsi, I.; Hannus, I.; Tihanyi, T.; Kiss, A. In Catalysis by Zeolites; Imelik, B., et al., Eds.; Elsevier: Amsterdam, 1980; pp 135140. (19) Kiricsi, I.; Hannus, I.; Kiss, A,; Fejes, P. Zeolites 1982, 2, 247. (20) Park, Y. S.; Lee, Y. S.; Yoon, K. B. J . Am. Chem. SOC.1993,115, 12220. (21) Anderson, P. A.; Barr, D.; Edwards, P. P. Angew. Chem. 1991, 103, 15ll;Angew. Chem., Int. Ed. Engl. 1991, 30, 1501. (22) Srdanov, V. I.; Haug, K.; Metiu, H.; Stucky, G. D. J . Phys. Chem. 1992, 96, 9039. (23) Lui, X.; Thomas, J. K. Langmuir 1992, 8, 1750. (24) Lui, X.; Thomas, J. K. Chem. Phys. Lett. 1992, 5, 6, 555. (25) Abou-Kais, A,; Vedrine, J. C.; Massardier, J. J . Chem. Soc., Faraday Trans. I 1974, 71, 1697. (26) Abou-Kais, A,; Vedrine, J. C.; Massadier, J.; Dalmai-Imelik, G. J. Chem. Soc., Faraday Trans. 1 1973, 70, 1039. (27) Sinfelt, J. H. Bimetallic Clusters; John Wiley & Sons: New York, 1983. (28) Sinfelt, J. H. Science 1977, 195, 641. (29) Simon, M. W.; Efstathiou, A. M.; Bennett, C. 0.; Suib, S. L. J . Catal. 1992, 138, 1. (30) Xu, B.; Chen, X.; Kevan, L. J . Chem. Soc., Faraday Trans. 1991, 87, 3157. (31) Guy, S. C.; Edwards, P. P. Chem. Phys. Lett. 1982, 86, 150. (32) Walsh, W. M.; Rupp, L. W.; Schmidt, P. H. Phys. Rev. Lett. 1981, 16, 181. (33) Anderson, P. A.; Edwards, P. P. J . Am. Chem. SOC. 1992, 114, 10608. (34) Vedrine, J. C.; Abou-Kais, A,; Massardier, J.; Dalmai-Imelik, G. J . Catal. 1973, 29, 120. (35) Kasai, P. H.; Bishop, R. J. J . Phys. Chem. 1973, 77, 2308. (36) Kaisi, P. H.; Bishop, R. J. Adv. Chem. Ser. 1976, 171, 350. (37) Pfeffer, R. C. J . Appl. Phys. 1985, 57, 5176. (38) Suib, S. L.; Morse, B. E. Langmuir 1989, 5, 1340. (39) Shih, S. J . Catal. 1983, 79, 390. (40) Vincow, G. Radical Ions; Interscience: New York, 1968. (41) Handbook of Chemistry and Physics, 64th ed.; CRC Press: Boca Raton, FL, 1983; p E-71.
J. Phys. Chem., Vol. 99, No. 13, 1995 4709 (42) Shida, T.; Egawa, Y.; Kubodera, H. J . Chem. Phys. 1980, 73,5963. Shih, S. J . Phys. Chem. 1975, 79, 2201. (43) Corio, P. L.; Shih, S. J . Catal. 1970, 18, 126. (44) Ratcliffe, C. I.; Ripmeester, J. A.; Tse, J. S. Chem. Phys. Lett. 1985, 120, 427. (45) Hunger, M.; Freude, D.; Pfeifer, H. Catal. Today 1988, 3, 507. (46) Hayashi, S.; Hayamizu, K.; Yamamoto, 0. Bull.. Chem. SOC.Jpn. 1987, 60, 105. (47) McMunay, L.; Holmes, A. J.; Kuperman, A,; Ozwi, G. A,; Ozkar, S . J . Phys. Chem. 1991, 95, 9448. (48) Simon, M. W.; DeGuzman, R. N.; Suib, S. L. Unpublished results. (49) Efstathiou, A. M.; Borgstedt, E. v. R.; Suib, S. L.; Bennett, C. 0. J. Catal. 1992, 1.35, 135. (50) Ward, J. W. In Zeolite Chemistry & Catalysis; Rabo, J. A,, Ed.; American Chemical Society: Washington, DC, 1976; p 187. (51) Pesek, J. J.; Sandoval, J. E.; Su, M. J . Chromatogr. 1993, 690, 95. (52) Fraser, G. W.; Greenwood, N. N.; Straughan, B. P. J . Chem. SOC. 1963, 3742. (53) Jacobs, P. A. Carbogenionic Activity of Zeolites; Elsevier: Amsterdam, 1977. (54) Noumi, H.; Misumi, T.; Tsuchiya, S. Chem. Lett. 1978, 493. (55) Martens, L. R.; Vermeiren, W. J.; Huybrechts, D. R.; Grobet, P. J.; Jacobs, P. A. In Proceedings, 9th Intemational Congress on Catalysis, Calgary, 1988; Phillips, M. J., Teman, M., Eds.; The Chemical Institute of Canada: Ottawa, 1988; Vol. 1, p 420. (56) Roginski, S. Z.; Rathmann, J. J . Am. Chem. Soc. 1933, 55, 2800. (57) Lee, J. B. J . Catal. 1981, 68, 27. (58) Bonnevoiot, L.; Olivier, D.; Che, M. J . Mol. Catal. 1983,21, 415. (59) Muha, G. M. J . Phys. Chem. 1966, 70, 1390. (60) Kohn, H. W.; Taylor, E. H. Acres du 2eme Congres Catalyse; Technip: Paris, 1960; p 1461. (61) Hays, J. D.; Bunbar, R. C. J . Phys. Chem. 1979, 83, 3183. (62) Vedrine, J. C.; Barthomeuf, D.; Dalmai, G.; Trambouze, Y.; Imelk, B.; Prettre, M. Compt. Rend. C 1968, 267, 118. JP94285 1X
